Recently, there has been a growing need for a reliable prediction of the effect of psychiatric treatment for (partially) hospitalized patients. Prediction research, however, has until now mainly been aimed at the determination of contra-indications for treatment, instead of predictors for treatment success. In this article an exploratory prediction study is described, based on a sample of 150 clinical and day-clinical psychiatric patients. The aim of the study was: which patient characteristics serve as predictors for treatment success?

Variables used were demographic and admission data, psychiatric history, psychopathology, social and interpersonal functioning, social network and social support. The sample (n = 150) was divided into two equal groups, a relatively more successful and a relatively less successful group.

Logistic regression analysis revealed that group membership could be predicted correctly for 76% of the cases. Nine predictors were used, including interactions for eight of them. Among those predictors were the period between referral and admission, social network variables, and psychiatric diagnosis. However, the gravity of psychiatric symptoms and of social dysfunctioning appeared to be no predictors for treatment success.